13 dimensions, 12 weighted, one signal beside them.
Twelve weighted dimensions resolve into the two sub-indices, blended into the headline at the default split of 60% Sub-index A and 40% Sub-index B. Return Risk is reported beside the Index at weight zero. Weights below are the v8.0 published defaults; each column sums to 100% within its sub-index.
What Sub-index B is measured against.
Sub-index B does not score in the abstract. Each of its dimensions is graded against the live requirements of the surfaces that now route machine-driven demand: the marketplace feeds a product must qualify for, and the agentic protocols an agent transacts through.
Feed Protocol Compliance scores against the feeds
The marketplace and shopping-feed requirements that decide eligibility and rank. We score your catalog against what each feed demands.
Structured Data, Agent Required Attributes, and Natural-Language Parseability score agent readiness
Readiness against the agentic commerce protocols that route machine-driven demand: the Agentic Commerce Protocol (ACP) behind ChatGPT checkout, and Google's Universal Commerce Protocol (UCP) behind AI Mode and Gemini. These specifications change constantly; the score reflects current requirements.
What the score is built on.
Classification standard for technical products, used in distribution. Leads on B2B and HVAC/R.
B2B classification and product-description standard for technical and industrial products.
Global product and service classification standard.
GDSN plus GS1 identifiers, including GTINs: the global standards for product data and identification.
The Google Product Taxonomy, the category structure for shopping surfaces.
Marketplace product-data requirements, leading on consumer catalogs.
Honesty about confidence is part of the design, not a caveat bolted on.
The methodology grades the strength of evidence behind each measure and flags provisional ones until they are calibrated against a scored corpus. A score is always traceable to the evidence behind it.
Version-tracked, and open to challenge.
A methodology you cannot track over time is one you cannot trust. Every change to the dimensions, the weights, the standards, or the rules is versioned and recorded in a public change log. If you think a weight is wrong or a standard is misapplied, there is a way to say so, and a process that acts on it.
Challenge the methodology.
Tell us which weight, standard, or dimension you would change, and the argument for it. Substantive challenges are reviewed in the calibration cycle and credited in the change log.
Submit a methodology challenge →Questions about the methodology.
Why publish the methodology openly?
Because independence only means something if it is auditable. If we kept the scoring private, "independent" would be a claim. Publishing the dimensions, the weights, and the standards we score against makes it a fact you can check.
Can I comment on or challenge the methodology?
Yes. The methodology is a living document, and active evaluations and comments are invited to shape the next standards release. If you think a dimension is weighted wrong or a standard is misapplied, that input is wanted, not just tolerated.
How often does the methodology change?
It is version-tracked, with a published version label and a change history. Changes are deliberate and documented rather than silent, so a score is always traceable to the version of the methodology that produced it.
What standards does the methodology build on?
The active standards are ETIM, ECLASS, UNSPSC, GS1, the Google Product Taxonomy, and Amazon's requirements. The methodology page documents how each is used and weighted.
Is the score provisional or final?
The methodology marks measures as provisional until they are calibrated against a scored corpus, and discloses the strength of evidence behind each. That honesty about confidence is part of the design, not a caveat bolted on.